[Webkit-unassigned] [Bug 66295] [Meta] Support w3c reftests

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=66295


Peter Linss <peter.linss at hp.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |peter.linss at hp.com




--- Comment #28 from Peter Linss <peter.linss at hp.com>  2011-11-03 12:45:56 PST ---
(In reply to comment #27)
> (In reply to comment #24)
> > According to Peter Linss, the naming conventions are slightly more complicated. His test harness checks for (at least) filename is: *-ref.* ref-*.* (or *-notref.*  notref-*.*) and whether the file is in a "reftest" folder. He also ends up reading through some files and if there is not a rel=help link then he concludes the file is not a test.
> 
> This sounds like a serious problem to me. Can we wait until W3C converges on one method to write tests. It appears that people are already confused by the current implementation of ref tests and adding multiple ways to write ref tests will worsen the issue.

Let me be clear. We have one preferred method to name test files: append -ref or -notref to the filename of the reference file. 

Our system optionally supports ref- or notref- prefixes, or having the reference files in a 'reftest' or 'reference' sub-directory. These are for compatibility with older tests.

We also expect all retests to use the <link rel='match' .../> or <link rel='mismatch' .../> convention for source files. However our test build system (not the harness) will also generate reftest manifest files so it's not necessary for any other tools to parse the source files looking for reference links.

In general we expect testers to run the output of our build process, not the source.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.